Notion’s Token Town: 5 Rebuilds, 100+ Tools, MCP vs CLIs and the Software Factory Future — Simon Last & Sarah Sachs of Notion

2026-04-07 · Source: Latent.Space - Www.latent.space · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Expert, extended

Summary

Notion's cofounder and head of AI discussed the recent launch and development journey of their Knowledge Work AI agents, emphasizing a slow, iterative shipping process that involved rebuilding the agent system multiple times since late 2022. Key challenges included the immaturity of early LLMs and the absence of robust function calling, which necessitated Notion's internal development of tool-calling frameworks. The company prioritizes a "portfolio approach" to product development, balancing AGI-pilled projects with immediately useful features. Notion's strategy focuses on understanding user collaboration needs and building a platform where agents can seamlessly integrate, rather than solely developing frontier AI models. They also detailed their unique engineering culture, which encourages low ego, rapid prototyping, and a willingness to delete and rebuild code based on evolving capabilities and user feedback, leading to successful launches and high user adoption rates.

Key takeaway

For AI Engineers and ML Directors evaluating agentic system development, Notion's experience highlights the importance of designing for composability and user-centric workflows. Your teams should focus on building robust, adaptable harnesses that can integrate evolving model capabilities and diverse tools, rather than solely chasing frontier model training. Prioritize clear product conviction and a culture that embraces iterative rebuilding, as this directly impacts velocity and the ability to deliver valuable, reliable AI solutions that enhance existing collaboration platforms.

Key insights

Notion's AI development prioritizes user collaboration and iterative refinement over raw model capabilities, adapting to evolving LLM features.

Principles

• Prioritize user needs over cool tools.
• Give models what they want (e.g., Markdown, SQLite).
• Foster a low-ego, adaptable engineering culture.

Method

Notion's development involves rapid prototyping, internal dogfooding, and a "demos over memos" approach, with a strong focus on user journeys and continuous iteration based on model capabilities and user feedback.

In practice

• Use internal hackathons to uplift general population.
• Implement manager agents to abstract notifications.
• Treat memory as pages/databases for agents.

Topics

• Notion AI Agents
• Custom Agents
• MCP vs CLI
• Software Factory
• AI Evals & MBEs

Best for: AI Engineer, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• Why Notion’s biggest AI breakthrough came from simplifying everything — Simplifying AI interactions with human-readable formats and minimal abstraction dramatically improves model performance and user adoption.
• CLI vs MCP: How AI Agents Choose the Right Tool for the Job — CLI excels for known commands, while MCP provides necessary abstraction and governance for complex tasks.
• 🔵 Notion's new Workers can build Stripe Dashboards and is a Claude Backlash brewing? — AI agents and platforms are transforming product development and user interaction through automation and proactive capabilities.
• The Real Lesson from OpenAI’s Top Customers: Tokens Aren’t Spend. They’re Leverage — AI is becoming embedded infrastructure, automating cognition-heavy work across diverse sectors and roles, leveling the competitive landscape.
• From Coder to Manager: Navigating the Shift to Agentic Engineering with Notion Co-Founder Simon Last — Notion is evolving into an agent-centric platform, shifting from human-direct work to human-managed AI agents.
• DataRobot Q4 update: driving success across the full agentic AI lifecycle — Transitioning AI agents from prototype to production requires robust orchestration, governance, and scalable infrastructure.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Latent.Space - Www.latent.space.